Ground Balance Theory

[Ferric Toes]

I suggest this one which is used in the QED and can be used as a log amp,very low noise and fast slew rate.(very expensive though)
dougAEGPF
http://www.ti.com/lit/ds/symlink/vca810.pdf

Thanks. I have some VCA610's that I had forgotten about. These are much the same as VCA810 and I will resurrect them and see how they work as log amp in the feedback path of LME49990.

Eric.

[Aziz]

You guys causing me a lot of work. Yes, indeed. So I have to bite the bullet and have to implement the decay functions and show you the FFT response. Today is a nice weather. But monday/tuesday.
Stay tuned..
Aziz

[mikebg]

I suggest this one which is used in the QED and can be used as a log amp,very low noise and fast slew rate.(very expensive though)

Doug, there is a very cheap solution - a capacitor only connected in feedback.
Frequency analysis of received signals shows that it is better the preamp of QED to have linear decreasing frequency response instead log amplification:

http://www.geotech1.com/forums/showthread.php?20238-PI-preamp-frequency-response&p=167526#post167526

http://www.findmall.com/read.php?34,1550930,page=1

[dougAEGPF]

Anybody know of a good low noise log amp? I suspect that a bit of linear low noise preamplification ahead of it would be a good idea.

Eric.

I suggest this one which is used in the QED and can be used as a log amp,very low noise and fast slew rate.(very expensive though)
dougAEGPF

http://www.ti.com/lit/ds/symlink/vca810.pdf

[Midas]

This is what a log display should look like. Taken by Tektronix scope with a plug in waveform scanner outputed to XY plotter. Tube technology too - not bad. Log amp was by Optical Electronics and was 8 pin DIL - also not bad for 1968.
You can see that the rings 1 - 4 only differ in lateral width and display a single time constant. 5 is a disc and 6 is a 2in thick piece of bar. 5 shows a bit of curvature at early time and 6 curvature to around 400uS indicating a long TC that has not reached its fundamental TC until that time. The big spike at 400uS is an artifact and other spikes are noise.

I have a plot for different coins which shows nicely how each coin can be identified by its slope. Cupro nickel being fast copper and silver less slow. If you differentiate these waveforms you get a constant level which becomes noisier at later times. All except 5 and 6 where the differentiated waveform will change until the fundarmental TC is reached.

Iron and magnetic soil are quite different and still look exponential unless you have a log time plot.

It may look exponential but its not. So even after you plot it on a log plot (so taking the log twice) its still not a straight line (though much close to it compared to the original signal) just another even weaker 1/t^a response.

I can see the value in the log-amp for making the process clearer to humans. But the information is still there with a regular amp and I'm not convinced the log amp will match it in SNR. After all the log amp is amplifying least when the signal is strongest... I know as the signal gets weaker the amplification should rise.. but it still seems a bit non-ideal. In a sense your throwing away good signal by not amplifying it as much as possible when its at its best.

As far as identifying coins, does the orientation of the coin not have a big effect on its TC ? If it can't identify a coin regardless of how it sits in the ground it's not much help.

BTW: Thanks for the info on low delay sampling a while back.
Midas

[Ferric Toes]

Yes , you are right - I found this coil wire eddy current signal when I started my experiments with recuperative PI . In the classic PI we cannot notice it because it's being masked by the main process of the coil discharge ( LR exponent ) . I call this "the coil feels itself" This is why I use only a Litz wire to wind the coils in my new experimental devices . But even a Litz wire has this issue , of course less in magnitude .... Of course it's a problem , but not very big - because this parasitic signal is relatively weak , and we can reject it by a simple RC network , inserted in the preamp circuit .

Better solution is to add a little current transformer , resistor loaded , with time constant equal to the parasitic eddy current TC , and subtract its signal ( eddy current simulation ) from the main coil signal - so we can obtain almost ideal horizontal line on the scope without metal near the coil . I tried this , and it works perfect . So when we bring the metal to the coil , we can see only the pure target signal , without any distortions . Of course , it allows more reliable further signal analysis , and as I think , it will help to completely solve this ground signal problem .

I found that capacitive overshoot was the problem, particularly if there was any significant length of cable. I used to counter this with a piece of lead foil trimmed with scissors till it had a TC exactly that of the overshoot (undershoot if the preamp was non-inverting). The piece of lead foil was placed in the coil shell in such as position that it matched the overshoot amplitude. It was then potted in with the coil winding. Result - nice flat response with minimum damping. Actually, of course, the lead foil is acting as a small inductor and resistor coupled to the main coil.

Eric.

[Ferric Toes]

Interesting curves. What type and size of coil is used for testing, relative to target size, and what approx. log. vertical scale is, also if any signal subtraction is performed in the process, target-no target condition? I tried with 10in mono, otherwise very fast, and over 80dB dynamic range behind and failed to notice any significant change in decay slope. Actually, decay slope was even less affected by metal target compared to linear high gain amp, by observing just waveform. Maybe i have to retest this.

Whether I can find my notes is debatable having had many moves in between, Coil was probably 6in D at most. Vertical scale probably 3 decades. This is raw data straight off the preamp with no processing. You should clearly see changes in slope for different objects as in the above graphs.

Eric.

[deemon]

In this “initial decay” component, one part cannot be canceled out, it is not produced by LCR decay or some energy recovery mechanism, but due to finite eddy current decay in coil itself, target signal is then just superimposed to it. This component can be minimized to some degree using various tricks, but cannot be eliminated completely, using practical coils (mono). Log amp will just limit ability to distinguish between this and target influence, very small variation will now be displayed in log scale. Exp or sq. can potentially generate more useful difference information. No flat lines here. Potential discrimination is then target TC imposed on coil TC signal. I measured this on actual hardware, not simulated or calculated.

Yes , you are right - I found this coil wire eddy current signal when I started my experiments with recuperative PI . In the classic PI we cannot notice it because it's being masked by the main process of the coil discharge ( LR exponent ) . I call this "the coil feels itself" This is why I use only a Litz wire to wind the coils in my new experimental devices . But even a Litz wire has this issue , of course less in magnitude .... Of course it's a problem , but not very big - because this parasitic signal is relatively weak , and we can reject it by a simple RC network , inserted in the preamp circuit .

Better solution is to add a little current transformer , resistor loaded , with time constant equal to the parasitic eddy current TC , and subtract its signal ( eddy current simulation ) from the main coil signal - so we can obtain almost ideal horizontal line on the scope without metal near the coil . I tried this , and it works perfect . So when we bring the metal to the coil , we can see only the pure target signal , without any distortions . Of course , it allows more reliable further signal analysis , and as I think , it will help to completely solve this ground signal problem .

[Tepco]

Interesting curves. What type and size of coil is used for testing, relative to target size, and what approx. log. vertical scale is, also if any signal subtraction is performed in the process, target-no target condition? I tried with 10in mono, otherwise very fast, and over 80dB dynamic range behind and failed to notice any significant change in decay slope. Actually, decay slope was even less affected by metal target compared to linear high gain amp, by observing just waveform. Maybe i have to retest this.

[Ferric Toes]

Found the coins plot in a magazine article I wrote.

[Ferric Toes]

This is what a log display should look like. Taken by Tektronix scope with a plug in waveform scanner outputed to XY plotter. Tube technology too - not bad. Log amp was by Optical Electronics and was 8 pin DIL - also not bad for 1968.
You can see that the rings 1 - 4 only differ in lateral width and display a single time constant. 5 is a disc and 6 is a 2in thick piece of bar. 5 shows a bit of curvature at early time and 6 curvature to around 400uS indicating a long TC that has not reached its fundamental TC until that time. The big spike at 400uS is an artifact and other spikes are noise.

I have a plot for different coins which shows nicely how each coin can be identified by its slope. Cupro nickel being fast copper and silver less slow. If you differentiate these waveforms you get a constant level which becomes noisier at later times. All except 5 and 6 where the differentiated waveform will change until the fundarmental TC is reached.

Iron and magnetic soil are quite different and still look exponential unless you have a log time plot.

[Tepco]

This initial decay must be not enhanced but completely suppressed - it's the only way to good discrimination . I mean that we must have a zero signal without a metal target presence . So when we see the signal - it must be the target or ground signal , and it must be easier to separate them ...

In this “initial decay” component, one part cannot be canceled out, it is not produced by LCR decay or some energy recovery mechanism, but due to finite eddy current decay in coil itself, target signal is then just superimposed to it. This component can be minimized to some degree using various tricks, but cannot be eliminated completely, using practical coils (mono). Log amp will just limit ability to distinguish between this and target influence, very small variation will now be displayed in log scale. Exp or sq. can potentially generate more useful difference information. No flat lines here. Potential discrimination is then target TC imposed on coil TC signal. I measured this on actual hardware, not simulated or calculated.

[Tepco]

Never tried actual bricks, i have some very large ferrite for tests (E100 core) but i considered using one brick to disassemble test setup. On log scale presentation, notoriously small difference will be even smaller.

[Davor]

Actually i still have this hardware somewhere in my junkbox, maybe i can find and revive it and post some results.

Did you try bricks with that? Point is that in log domain targets of different tau will present themselves with unique slopes, shifted up or down due to proximity to the coil but at constant angle, so a mere derivation of these would give you taus directly. Magnetically relaxing materials would present themselves as ski-like curves, and in log time they straighten up. This could lead to their exclusion, hence GB. Otherwise it is irrelevant what kind of signal presentation you have, as long as it is not grazing on the noise floor.

[deemon]

Maybe exactly opposite approach may be better, some sort of anti-log amp, or squaring amp ( built using analog multiplier or something) to enhance initial decay part of waveform,

This initial decay must be not enhanced but completely suppressed - it's the only way to good discrimination . I mean that we must have a zero signal without a metal target presence . So when we see the signal - it must be the target or ground signal , and it must be easier to separate them ...